Connect with us
Nvidias new AI system translates text into landscape images Nvidias new AI system translates text into landscape images

Security

Nvidia’s new AI system translates text into landscape images

Published

on

A new system based on AI technology GauGAN2 allows you to create lifelike photos of nonexistent landscapes.

Nvidia presented a new system based on artificial intelligence technologies GauGAN2 (successor to the first model GauGAN), which allows you to create lifelike photos of nonexistent landscapes. Using techniques such as segmented mapping, retouching, and text-to-image conversion, the GauGAN2 is capable of creating lifelike images from text and freehand sketches.

“Compared to other advanced models, especially for converting text to image or map segments to image, the underlying neural network of GauGAN2 produces more varied and high-quality images. Instead of drawing each element of an imaginary image, users can simply enter a short phrase and generate it key features and plot like a snow-covered mountain range. This starting block can then be completed by making this or that mountain higher and adding trees in the background or clouds in the sky, “- said Nvidia team member Isha Salian.

GauGAN2 is an improved version of the GauGAN system, created in 2019 and trained on over a million open images from the Flickr platform. Like the GauGAN, the GauGAN2 understands the relationships between objects such as snow, trees, water, flowers, bushes, hills and mountains and “realizes” that the type of precipitation changes with the seasons.

Both GauGAN and GauGAN2 are a generative adversarial network (GAN) consisting of a generator and a discriminator. The generator takes samples (images with accompanying text) and assumes which data (words) correspond to other data (landscape elements). The generator is trained by tricking a discriminator that evaluates whether these assumptions are true. Although GAN hops are usually of poor quality, they are enhanced by discriminator feedback.

Unlike GauGAN, GauGAN2 is trained on 10 million images and is capable of translating speech descriptions into landscape images. If you enter text like “sunset on the beach”, the network will generate a corresponding image, and if you expand the phrase to “sunset on a rocky beach” or replace “sunset” with “noon” or “rainy day”, the image changes accordingly.

With GauGAN2, users can generate segmented maps – high-level sketches showing the location of objects in an image. This sketch can then be turned into a drawing by adding rough sketches to it using the sky, tree, rock, and river markings, or hand-painting with the brush tool.

Nvidia says the first version of GauGAN is already being used to create concept art for movies and video games. As with GauGAN, the company plans to release the GauGAN2 code on GitHub along with an interactive demo on the Playground, a web hub for Nvidia’s AI research and deep learning.

Click to comment

Leave a Reply

Your email address will not be published.

Security

Cryptocurrency exchange Binance was robbed of $570 million. Hacker withdrew BNB tokens

Published

on

Cryptocurrency exchange Binance was robbed of 570 million Hacker withdrew

Cryptocurrency exchange Binance has undergone another hacker attack. The attackers managed to withdraw BNB tokens worth about $570 million.

Cryptocurrency exchange Binance was robbed of $570 million.  Hacker withdrew BNB tokens

Somewhat earlier it was reported that the attack allowed the hackers to steal about $110 million, but now it turned out that everything is much worse. At the same time, the specialists of the exchange managed to freeze part of the funds, but we are talking about only 7 million dollars, which is clearly insignificant against the background of 570 million.

The exchange revealed that a cross-chain bridge connected to its BNB chain was attacked, allowing hackers to move BNB tokens off the network. Now the network has been restored, and the clients’ funds, according to Binance, are safe.

The fact of the theft of funds contributed to a sharp drop in the BNB rate by almost 5%, but after a few hours, almost half of the fall was redeemed.

Continue Reading

Security

Hacker who earned $27 million in cyberattacks will spend 20 years in prison and pay $21 million in fines

Published

on

Hacker who earned 27 million in cyberattacks will spend 20

A Florida district court has sentenced 34-year-old IT engineer Sebastien Vashon-Desjardins to 20 years in prison for carrying out at least 90 cyberattacks.

Hacker who earned $27 million in cyberattacks will spend 20 years in prison and pay $21 million in fines

It is noted that for several years of his activity, the hacker, using the NetWalker encryption virus, earned about $ 27 million. A search of Vashon-Desjardins revealed a crypto wallet containing 719 bitcoins, which was about $22 million at the time of the cybercriminal’s arrest in January 2022.

According to investigators, the 34-year-old cybercriminal acted in collusion with other hackers. Vashon-Desjardins himself played the role of an attacker: he infected the corporate networks of various companies with a virus and then demanded a ransom from them. Organizations from the USA, Canada and a number of European countries suffered from the activities of the criminal.

It is noted that, in addition to the prison term, the court also imposed a fine on Vashon-Desjardins in the amount of $ 21 million. Also, the criminal will have to pay compensation to the companies affected by his actions. The amount of damages has not yet been established.

Continue Reading

Components

Unique behavior of Ryzen 7000 processors. The notorious patches from the Specter vulnerability improve the performance of new CPUs

Published

on

Unique behavior of Ryzen 7000 processors The notorious patches from

Recently, various vulnerabilities in processors have been talked about much less often, and users no longer worry about performance degradation due to patches. As it turns out, Ryzen 7000 processors generally benefit from such patches!

Unique behavior of Ryzen 7000 processors. The notorious patches from the Specter vulnerability improve the performance of new CPUs

At least this is true for Linux, since it was in this OS that the author tested the Ryzen 9 7950X and Ryzen 5 7600X. It turned out that when working out of the box, the CPUs show better performance than when loading a special version of Linux with a deactivated patch from the Specter V2 vulnerability.

Unique behavior of Ryzen 7000 processors. The notorious patches from the Specter vulnerability improve the performance of new CPUs

Of course, such results do not appear everywhere, and during normal work they are unlikely to be critical. In particular, in total, according to the results of 190 tests, the difference was only 3%.

Continue Reading

Most Popular